[KongchangAI]
ConceptRisk-aware Orthogonal Subspace Navigation

ROSE

A framework for lifelong knowledge editing in multimodal large models. It employs orthogonal subspace constraints and risk masking to reduce interference between edits and catastrophic forgetting.

Timeline (last 90 days)

Nov 9

ROSE is a risk-aware orthogonal subspace navigation framework for lifelong knowledge editing in multimodal large models.

Unverified50%
Nov 9

ROSE restricts parameter updates to the orthogonal complement of previously edited subspaces, geometrically reducing mutual interference.

Unverified50%
Nov 9

The ROSE framework contains two key designs: orthogonal subspace navigation and a risk-guided mechanism, the latter dynamically computing risk masks to protect sensitive parameter subspaces.

Unverified50%
Nov 9

ROSE adopts a multi-granularity knowledge integration strategy during inference, including prototype-anchored semantic gating and perturbation-driven subspace fusion.

Unverified50%
Nov 9

The researchers conducted comprehensive experiments on ROSE across two mainstream multimodal benchmarks.

Unverified50%
Nov 9

ROSE is competitive with current state-of-the-art methods on five key metrics.

Unverified50%
Nov 9

ROSE maintains excellent stability even after more than 1,000 sequential updates.

Unverified50%
Nov 9

ROSE's code has been open-sourced at https://github.com/sunlitsong/ROSE.

Unverified50%

All Facts (8)

Source Articles